Harbor

A local LLM stack manager that connects Ollama, llama.cpp and vLLM to AI apps through Docker Compose, with a CLI and companion app.

Harbor helps people experiment with local LLMs by connecting model backends, chat interfaces and supporting services. Its CLI and companion app manage those connections in a shared development environment.

Models and chat interfaces connect backends such as Ollama, llama.cpp and vLLM to apps including Open WebUI and LibreChat.
Search, voice and images connect SearXNG to research tools such as Perplexica and Local Deep Research, Speaches to voice chat, and ComfyUI with FLUX to Open WebUI.
Coding assistants connect installed tools including Codex, Claude Code and OpenCode to local OpenAI-compatible backends; Harbor Boost adds workflows for research and checking agent deliverables.
Tools and workflows support MCP integrations and services such as Dify and n8n for building AI workflows.
Reusable setups let you save configuration profiles and export a selected stack as a standalone Docker Compose file.

Harbor is open source under Apache 2.0 and primarily manages services through Docker Compose. On macOS, Docker Model Runner, MLX and oMLX can run inference directly on the host with Metal acceleration. You can access its services from a phone over your local network.

Harbor is intended for local development and experimentation rather than production deployment. Web search and research features access external web services.